Methods: A systematic literature review was performed and all studies reporting frequency data on the association of a potential risk factor and procedural 30-day mortality of endovascular repair or open surgery in patients treated for ruptured AAAs were included. Pooled univariate effects from the literature data were calculated using random effects models to identify statistically significant preoperative risk factors (P<0.01). Subsequently, a prospective multicenter observational study was performed in which, at this point, 146 patients who underwent endovascular repair or open surgery for ruptured AAAs were included. Statistically significant preoperative risk factors as identified from the literature review were documented for each individual patient. An adaptation factor for each risk factor (i.e., difference in value between uni- and multivariable logistic regression coefficients in individual patient data) was calculated. Literature data and individual patient data were combined by adding the adaptation factor from the individual patient data to univariate regression coefficients as estimated from the literature. A prediction rule based on the combined data was developed. Prediction rule performance was tested in the prospective dataset by evaluation of calibration (plot observed versus predicted and Hosmer-Lemeshow test (p>0.20)) and discrimination (area under the ROC-curve°Ý0.8).
Results: Forty studies met the inclusion criteria and were included in the systematic review. Statistically significant risk factors in these studies were gender, age > 80 years, renal impairment, cardiac disease, and systolic blood pressure <90 mmHg on presentation. The adapted odds ratios (OR) included in the prediction rule were 2.14 for age > 80 years, 2.03 for renal impairment, 1.95 for cardiac disease, 3.03 for systolic blood pressure <90 mmHg, and 3.14 for the procedure performed (open surgery versus endovascular repair). The adapted OR for gender was 0.99; therefore gender was not included in the prediction rule. The prediction rule showed good calibration and discrimination.
Conclusion: The preoperative risk factors included in the prediction rule can be determined on patients' presentation to the hospital. This prediction rule predicts 30-day mortality after endovascular repair and open surgery in patients with ruptured AAAs.